bonded repair: design and process considerations · – reliance on bonds for flight safety...

1

Bonded Repair:Design and Process Considerations

Michael Borgman ~ Spirit AeroSystemsCo-Chair ~ Commercial Aircraft Composite Repair Committee

CACRC Spring 2007 ~ May 7-11 ~ Amsterdam

2

Introduction

• Generally, bonded repairs are not allowed in structurally critical applications

– Reliance on bonds for flight safety generally disallowed• One reason is strength performance scatter resulting from the somewhat limited precision inherent to bonded repair design, analysis, and processing

• A second is the absence of a means of verifying bond strength in the finished product

• This presentation provides thoughts on ways to enhance precision [of process control]

3

Introduction• Any scenario for process control must be evaluated for reliability of

success (probability must be assessed)• This reliability must be demonstrated through mechanical testing the

product from a LARGE volume of bondment process cycles• Have started this process at SPIRIT in support of one of our products,

and have tested scarf joint strengths from approximately 200 bondment process cycles

– Many different mechanics– Many different cure thermal profiles– Fresh and Old Material– Many different resulting NDI scan quality results

• Have learned there are nuances to be considered in selecting a joint geometry for process reliability testing

– Joint shape and edge proximity (breathing during cure)

4

Introduction

• This presentation is focused on bonded joints– Discussion topics

1. Regulations / Requirements2. Design and analysis

a) Joint type comparisonsb) Parametric trends

3. Surface preparationa) Geometry creationb) Pre-bond surface cleanlinessc) Pre-bond substrate moisture content

4. Laminate patch preparationa) Precision ply cutting

5. Post-bond cure state assessment

5

FAR Regulations• Design Requirements• 14 CFR 23.305 / 25.305 Strength and Deformation

– Support limit loads without detrimental permanent deformation– Support ultimate loads without failure for at least 3 seconds

• 14 CFR 25.571 Damage Tolerance and Fatigue Evaluation of Structure– Evaluation of strength, detail design, and fabrication must show

• Catastrophic failure due to fatigue, corrosion, manufacturing defects, or accidental damage, will be avoided

• Capability to successfully complete a flight during which likely structural damage occurs– [Impact, lightning (25.581), etc…]– Damaged structure able to withstand the static loads (considered as ultimate loads) which are reasonably

expected– Each evaluation must…

• [Include] typical loading spectra, temperatures, and humidities• [Be based on] analysis, supported by test evidence• [Make the assumption that] structure contains an initial flaw of the maximum probable size that

could exist as a result of manufacturing or service-induced damage

6

FAR Regulations• Design Requirements (cont’d)• 14 CFR 23.573 Damage tolerance and fatigue evaluation of structure(a) Composite Airframe Structure

– For any bonded joint, the failure of which would result in catastrophic loss of the airplane, the limit load capacity must be substantiated by one of the following

1. Maximum disbonds of each bonded joint consistent with the capability to withstand the loads in paragraph (a)(3) of this section must be determined by analysis, tests, or both. Disbonds of each bonded joint greater than this must be prevented by design features

2. Proof testing must be conducted on each production article 3. Repeatable and reliable non-destructive inspection techniques must be

established that ensure the strength of each joint

7

FAR Regulations• Design Requirements (cont’d)14 CFR 23.573 Damage tolerance and fatigue evaluation of structure(a) Composite Airframe Structure

– Demonstrate by tests, or by analysis supported by tests, structure capable of carrying ultimate load with damage up to the threshold of detectabilityconsidering the inspection procedures employed

– Growth rate or no-growth of damage from fatigue, corrosion, manufacturing flaws or impact damage, under repeated loads expected in service, must be established by tests or analysis supported by tests

– Requirements on residual strength with damage1. Critical limit flight loads with the combined effects of normal operating pressure and

expected external aerodynamic pressures 2. The expected external aerodynamic pressures in 1g flight combined with a cabin

differential pressure equal to 1.1 times the normal operating differential pressure without any other load

8

•“Scarf Lap Joint” (joint without load eccentricity)

•“Scarf Rate” = L / t

Joint GeometriesBasic Definitions: “Scarf Joint”

RepairSubstrate

Adhesive

L

tRepairSubstrate

Adhesive

L

t

Substrate

Substrate

Repair

Repair

9

•“Step Lap Joint” (joint w/moderate load eccentricity)

•“Step Rate” = L / t = “Per Ply Overlap” / “Ply Thickness”

Joint Geometries Basic Definitions: “Step Joint”

L

t

Substrate

Substrate

Patch

Patch

SubstratePatch

10

•Poor design details = “square cut” edges on damage removal and/or patch– Large bond stress riser at square edge

Joint Geometries Basic Definitions: Poor Joint Details

AA

AA

A-A

Substrate

PatchA-A

Substrate

Patch

11

Design Preface“Achilles Heel” Of Bonding

• Cannot verify bond strength via post-cure inspection• In practice, rigorous control of materials and processes is

approach used to ensure bond strength in final product• Post-cure NDI can identify some quality problems

– Porosity, delamination, inclusion, non-bond• However, “kissing bond” may not be found, and…• Fundamentally, the bond strength is unknown• As a result…

The structural integrity of any bond can be questioned

12

Design Preface“Achilles Heel” Of Bonding

• Absence of “NDI bond strength guarantee” limits scope of bonded repair applications

• These limits ARE necessary for today’s level of technology (if can’t prove it’s good, then must assume it’s not)

∴In response, the repair design philosophy is…Continued safe flight with repair completely failed

– Ultimate load capability restored by presence of repair– However, safe flight (limit) is not dependent on repair

13

Analysis PrefaceHistoric Perspective

• Analysis not generally performed on bonded repair joints• Instead, substantiation data is developed for specific joint geometry

(point design data)• The substantiated joint is then used without analysis• Substantiated joint generally chosen such that ultimate failure occurs

in adherends not in adhesive bond-line• Damage tolerance analysis not generally performed• “Point design” repair data approach has worked well for existing

relatively simple composites architectures• Next generation composite airframe structures are more complicated

architecture with more complex loading• Broader array of point design data will be required to cover the

breadth of potential application needs• May be impractical

14

Analysis Preface (continued)Approach Shift For Today’s Materials

• Today’s very strong adherend laminates sometimes require joint geometries with large joint overlaps (to fail adherends instead of adhesive bond)

• These large overlaps can sometimes significantly complicate repair geometry by impinging on surrounding structural details

• “Good” bonded joints display relatively repeatable failure stresses (peel/shear)

• From structural integrity perspective need predictable, repeatable, failure mode plus understanding of joint static strength, durability, and damage tolerance

• If sufficient data is created and if analysis method can be developed and proven effective ~ ”Is design for bond failure” a reasonable option to consider?

15

• General steps in bonded scarf repair (flush repair shown)1. Design and analyze repair2. Damage removal3. Prepare substrate repair geometry

• Scarf joint shown but other joint types exist4. Create patch (ply map)5. Clean surface6. Apply adhesive7. Lay-up patch8. Cure and inspect

Repair Process Steps

-45

-4545

900

090

45-45

-4545

900

090

45

-45

-4545

900

090

45-45

-4545

900

090

45

16

Repair “Joint” Analysis

17

Repair “Joint” AnalysisTaper Region Stiffness Decay

• Laminate stiffness decays in non-linear fashion in joint step or scarf (taper) region

• Linear stiffness variation only present if all plies same stiffness (i.e., uni-directional laminate)

• Analysis methods which neglect laminate stacking sequence cannot predict stress/strain distribution

18

Analysis Preface (continued)Taper Region Stiffness Decay

• Consider [90/45/0/-45/90/45/0/-45]s “all tape” laminate (25/50/25)

– Non-linear stiffness decay as “Step” taper is traversed10

0%

99%

95%

80%

75%

74%

70%

55%

50%

45%

30%

26%

25%

20%

5% 0.8%

0%20%40%60%80%

100%120%

16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1

Lam

inat

e St

iffne

ss

(lb/in

)

Ply Count

19

• Tapered “Scarf” joint displays similar behavior – Non-linear stiffness decay as “Scarf” taper is traversed

Analysis Preface (continued)Taper Region Stiffness Decay

20

Analysis Preface (continued) Adhesive Elastic-Plastic Behavior

• Epoxy Displays Elastic-Plastic Deformation– However, Behavior May Not Be Reliably Repeatable– Treating As Purely Elastic Sometimes Appropriate

Courtesy: NIAR

21

Analysis and DesignJoint Selection Considerations

• Adherend strain profile divided by far-field strain– Common adherends in all [90/45/0/-45/90/45/0/-45]s

• Step lap yields highest far-field strain multiplier– Adherend max strain ≈ 250% far-field strain

» Applies to specific configuration analyzed only

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

0

1

2

3

P

P

P

P

Tension

22

• Bond shear stress profile at equal load (normalized)– Same adherends in each case shown: laminate = [90/45/0/-45/90/45/0/-45]s

– Neglecting plasticity effects, stress 50% higher in step, 350% higher in square

0

1

2

3

4

Very large shear stressAt square edge detail

“Normal” value(Scarf joint peak shear )

Analysis and DesignJoint Selection Considerations

P

P

P

P

Tension

23

• Scarf rate effects– “Lion’s share” of benefit to adherends occurs prior to increase to 20:1– Bond shear stress “peak to valley” amplitude increases with increasing length

• Yields surprising fatigue behavior (fatigue sensitivity manifests sooner in longer lap)

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70 80 90Scarf Rate

MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain

Laminate = [90/45/0/-45/90/45/0/-45]s (Tape)

0.0

0.5

1.0

1.5

2.0

2.5

0 1Lap Position (x/L)

Loca

l Str

ess

/ Ave

rage

Str

ess

10:1 20:1 30:1

Analysis and DesignScarf Parametric Observations

P

P

P

P

Tension

24

• Scarf rate effects & damage tolerance considerations

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70 80 90Scarf Rate


0

1

2

3

4

5

6

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00x/L

Baseline ~ Bond Shr / AVE Bond ShrBond Shr / lb. Applied LoadBond Shr / AVE Bond ShrAdherend Strain / Far Field Adherend Strain

Approx. 300% Increase Analytic Residual Str. Approx• 30:1 Scarf Joint• With Impact Damage• 50% Bond Loss (Center)• Implies 2/3 Strength Loss• Lap Length Consideration

For Damage Tolerance

Laminates[90/45/0/-45/90/45/0/-45]s (Tape)


P

P

P

P

Tension

25

• Adhesive elastic modulus effects– Significant effect on adherend strain and peak-to-valley shear

stress in range of typical epoxy adhesives

0

1

2

3

4

0 50 100 150 200Adhesive Modulus (ksi)


Basis• 30:1 Scarf Rate• Constant Adhesive Thickness

0

1

2

3

4

0 50 100 150 200Adhesive Modulus (ksi)


Basis• 30:1 Scarf Rate• Constant Adhesive Thickness

Typical Epoxies

Laminates[90/45/0/-45/90/45/0/-45]s (Tape)


P

P

P

P

Tension

26

• Adhesive thickness effects– Significant effect on adherend strain and peak-to-valley shear stress in range

of typical epoxy film adhesives (for single ply)

0

1

2

3

0.000 0.009 0.018 0.027 0.036 0.045 0.054Adhesive Thickness (in.)


Basis• 30:1 Scarf Rate• Constant Adhesive Modulus

Typical Epoxy Film Adhesives

0

1

2

3

0.000 0.009 0.018 0.027 0.036 0.045 0.054Adhesive Thickness (in.)


Basis• 30:1 Scarf Rate• Constant Adhesive Modulus

Typical Epoxy Film Adhesives

Laminates[90/45/0/-45/90/45/0/-45]s (Tape)


P

P

P

P

Tension

27

• Laminate stacking sequence (16 ply laminates)– Interchange 0º and 90º plies

• Increased max. bond stress, Decreased max adherend strainBond ~ [90/45/0/-45/90/45/0/-45]sBond ~ [0/45/90/-45/0/45/90/-45]s

Bond Shear Stress Profile

Substrate ~ [90/45/0/-45/90/45/0/-45]sSubstrate ~ [0/45/90/-45/0/45/90/-45]s

Adherend Axial Strain Profile


P

P

P

P

Tension

28

Side Note: Current FAA WorkAnalysis / Damage Tolerance

• FAA collaborative research in process at NIAR• Scope: 1) analysis methods development, 2) bond

process robustness assessment, and 3) repair damage response

• Four major thrusts1. Baseline mechanical performance

– Static strength / post-fatigue residual strength / durability2. Degradation of mechanical performance as result of

contaminants on bond surface3. Impact damage site [in joint] resulting in greatest performance

degradation (i.e., What is critical site?)4. Detect-ability of bond surface contaminants

– Adequate pre-bond surface preparation / cleaning

Parent

Repair

Parent

Repair

Parent

Repair

Parent

Repair

29

Process Considerations

Surface Preparation

30

Surface PreparationScarf Grinding Operations

• Damages seldom occur in region with constant laminate thickness

• Ply drops in region effect final taper sand geometry• Sometimes difficult to anticipate what the final material

removal product should look like• Beneficial to have visualization of final taper sand prior to

starting grinding operations• During initial design, many airframe components modeled as

3-D entities in space and on A ply-by-ply basis• With appropriate software can generate 3-D simulation of

taper sanded region (prior to start)

31

Surface PreparationScarf Grinding Operations

• Three dimensional simulation enhances ability to produce repairs of higher precision

• Provides greater confidence that sanding operation was performed correctly

– Mechanic visual aid– Data for mechanizing major material removal steps

using robotic grinding• Hand finishing still required

– Improved records demonstrating that final repair geometry “looked like it was supposed to”

32

Surface PreparationPrecision Scarf Grinding

• Automated grinding using digital model data

Repair PliesDoubler Plies

Repair PliesDoubler Plies

33

Surface PreparationPre-Bond Surface Inspection

“Compact instrumentation” inspections methods– Contact angle (surface energy resulting from contaminants)– Near IR diffuse reflectance spectroscopy (shown)

Courtesy: Wichita State University, Department of Chemistry (FAA Collaboration)

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

40005000600070008000900010000

Abs

orba

nce

Wavenumbers (cm-1)

Synthesized “934” Resin

R2 = 0.9919

0

1

2

3

4

5

6

0 1 2 3 4 5 6

Actual

Cal

cula

ted

Water Uptake

( p )

R2 = 0.9882

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0.5 1.5 2.5 3.5 4.5 5.5

Actual

Cal

cula

ted

Water Desorption

34


Laminate Patch Preparation

35

Process ConsiderationsLaminate Patch Preparation

• Even in repair scenarios with simple substrate geometry the patch plies seldom have regular, well defined edges

• A ply kit is created by….1. Taping clear Mylar over taper sanded region2. Hand tracing ply boundaries3. Transferring ply definition from Mylar4. Hand cutting repair plies with snips

• Final patch fit is marginal at best• The process is slow, and hand cutting

plies difficult to accurately accomplish• Randomness of patch fit adds to

randomness of structural performance

Example of Simple Taper Sand

36

Process ConsiderationsPrecision Laminate Patch Preparation

• Structural performance benefit in precision patch “ply kit”– Reduce randomness of structural performance

• Digital image processing is candidate technology to accomplish this

• With appropriate lighting, individual ply orientations can be detected using AFI instrumentation (Spirit AeroSystems patent pending)

• Points of prescribed density can be automatically defined in orientation change demarcation zones

• Splines laid through points to create ply edges• Feed data to automatic knife for precision cutting• Feed data to laser projector for patch ply lay-up• Provides permanent digital record of patch accuracy

37


• Variation of lighting accentuates ply orientations

Spirit Aerosystems ~ Patent Pending

38


• Examples: optical edge identification & digitization– Production of high precision “patch ply kits” possible

Spirit Aerosystems ~ Patent Pending

39


Cure State Verification

40

Post Bond Cure State

• Resins are uniquely individual with regards to cure response to thermal cycle

– Some resins achieve an approximately full cure state early in thermal cycle

– In other resins cure reaction initiates late in cycle• Repairs frequently cured using heat blankets

– Heat blankets deliver thermal profile with tolerance band broader than autoclave cure

– Some areas hotter than others• Important to verify full cure

• DOT/FAA/AR-03/74, Office of Aviation Research, “Bonded Repair of Aircraft Composite Sandwich Structures”, John S. Tomblin, Lamia Salah, John M. Welch, Michael D. Borgman, 2004

• http://www.tc.faa.gov/its/worldpac/techrpt/ar03-74.pdf

41

Post Bond Cure State Inspection

• Hand held diffuse reflectance infrared spectroscopy– For assessment of cure state by peak amplitude area

11

Antares II

Remote Triggered Diffuse Reflectance Module

Integrating Sphere Reflectance Module

11

Antares II

Remote Triggered Diffuse Reflectance Module

Integrating Sphere Reflectance Module


42


• Cure state by diffuse reflectance IR spectroscopyCuring FM377U adhesive

0.00

0.20

0.40

0.60

0.80

1.00

400042004400460048005000

uncured41% cure98% cure

Wavenumbers (cm-1)

Abs

orba

nce

Curing FM377U adhesive

0.00

0.20

0.40

0.60

0.80

1.00

400042004400460048005000


Wavenumbers (cm-1)

Abs

orba

nce

0.00

0.20

0.40

0.60

0.80

1.00

400042004400460048005000


Wavenumbers (cm-1)

Abs

orba

nce

0.00

0.20

0.40

0.60

0.80

1.00

400042004400460048005000


Wavenumbers (cm-1)

0.00

0.20

0.40

0.60

0.80

1.00

400042004400460048005000


Wavenumbers (cm-1)

Abs

orba

nce

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)

Nor

mal

ized

pea

k ar

ea

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)

Nor

mal

ized

pea

k ar

ea


Chemometrics Calibration Curve for cure of FM377U adhesive

R2 = 0.987

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2Actual % cure

Cal

cula

ted

% c

ure

43


• Cure state by diffuse reflectance IR spectroscopyCuring T300/934 prepreg

0.00

0.20

0.40

0.60

0.80

1.00

1.20

400042004400460048005000

uncured42% cured94% cured

Wavenumbers (cm-1)

Abs

orba

nce

Curing T300/934 prepreg

0.00

0.20

0.40

0.60

0.80

1.00

1.20

400042004400460048005000


Wavenumbers (cm-1)

Abs

orba

nce

0.00

0.20

0.40

0.60

0.80

1.00

1.20

400042004400460048005000


Wavenumbers (cm-1)

0.00

0.20

0.40

0.60

0.80

1.00

1.20

400042004400460048005000


Wavenumbers (cm-1)

0.00

0.20

0.40

0.60

0.80

1.00

1.20

400042004400460048005000


Wavenumbers (cm-1)

Abs

orba

nce


Chemometrics Calibration Curve for cure of T300/934 prepreg

R2 = 0.983

35

45

55

65

75

85

95

105

35 45 55 65 75 85 95 105

actual % cure

Cal

cula

ted

% c

ure

44

Summary of Observations• For structurally critical bonded repairs to be pursued, absence of NDI

method for verifying bond strength creates need for high precision repairs using highly constrained processes performed by highly trained personnel (mechanics and analysts)

– Proof of statistical reliability only existing alternative for strength guarantee• Reliability must be demonstrated in terms of analysis & design,

surface preparation, patch “fit” precision, and process validation & records

• Bond-line failure mode may be consideration (or requirement) in lieu of traditional requirement for adherend failure mode

– Fairly repeatable / predictable mode (given adequate repair precision)– However, need A great deal more substantiation data to verify it as

reasonable alternative

45

Summary of Observations• Joint type strongly impacts peak bond stresses• Joint geometry parameters must be chosen with care

– Some parameters are “double edge sword”– For example scarf rate can improve static strength but, at the same time,

reduce fatigue resistance (increase sensitivity to fatigue)• Analysis methods must include laminate stacking sequence effects

– Including stiffness decay in taper region• Research on performance of damaged scarf joints is needed

– Propagation characteristics– Damage containment feature development and demonstration

• Candidate precision repair technologies exist but need maturation• Moisture detectable by diffuse reflectance near IR spectroscopy• Cure state detectable by diffuse reflectance near IR spectroscopy

46

Future• In any process control scenario reliability of success will

have to be proven• This reliability must be demonstrated through mechanical

testing the product from a LARGE volume of bondmentprocess cycles

• I have started this process at SPIRIT in support of one of our products, and have tested scarf joint strengths from approximately 200 bondment process cycles

– Many different mechanics– Many different cure thermal profiles– Many different resulting NDI scan quality results

• Have learned there are nuances to be considered in selecting a joint geometry for process reliability testing

– Joint shape and edge proximity (breathing during cure)

47

END

bonded repair: design and process considerations · – reliance on bonds for flight safety...

Documents